INDEX

Explanations

terms related to coding, data management, and operational practices

oai_token-act-pair · gpt-3.5-turbo

references to various types of practices, particularly highlighting those that are problematic or questionable

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.6.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ergy

-0.71

joy

-0.70

otos

-0.70

pad

-0.69

amaz

-0.69

parts

-0.65

plane

-0.65

amen

-0.64

proof

-0.63

Vel

-0.63

POSITIVE LOGITS

hops

1.05

afety

0.95

 governing

0.89

pring

0.82

 affecting

0.80

etter

0.80

 inherent

0.79

ystem

0.78

uits

0.77

uggest

0.77

Activations Density 0.100%

terms related to coding, data management, and operational practices

references to various types of practices, particularly highlighting those that are problematic or questionable

No Comments

No Known Activations